Comparison of two approaches to standardize catch-per-unit-effort for targeting behaviour in a multispecies hand-line fishery

Abundance indices are difficult to derive from multispecies catch-per-unit-effort (CPUE) data as these critically rely on the ability to account for targeting behaviour that directs effort towards a particular target species or species-complex. We developed a novel standardization method that uses continuous principal component scores, derived from a Principal Component Analysis of the catch composition data, as nonlinear predictor variables for targeted effort within a Generalized Additive Model (GAM) framework. This 'Direct Principal Component' method (DPC) is compared to a commonly used standardization approach, in which clusters of fishing tactics are identified from the catch composition and treated as categorical variables. We employed GAMs, which incorporate spatial, seasonal, and fishing power effects, to evaluate the statistical performance of the two alternative sets of covariates. The GAMs were applied to CPUE datasets from the South African hand-line fishery, covering three fishing regions over a period of 26 years, to provide abundance indices for two commercially important species, carpenter, Argyrozona argyrozona and silver kob, Argyrosomus inodorus. Results show that both approaches removed substantial variation from the CPUE data, but that the DPC approach was statistically superior as judged by the AIC, deviance explained and results from 10-fold bootstrap cross-validations. The DPC method represents a more direct approach, is less time-consuming and subjective and therefore considerably easier to implement. We present evidence that the DPC was able to account for substantial shifts in targeting without masking the year effect. The standardized CPUE time series showed positive responses for both species as total landings decreased due to forced effort reductions in 2000. The approach presented here has potentially unlocked useful information for stock assessment in the over two million individual catch records available for the South African hand-line fishery and similar multispecies fisheries. (C) 2012 Elsevier B.V. All rights reserved.